What evidence supports a sales forecast?

This matters when hiring, capacity or delivery commitments depend on expected sales. Two forecasts with the same total may rest on very different evidence and timing assumptions.

Define what is being forecast

An order being placed, a product being delivered, an invoice being issued and a payment being received are different events. Specify which event “sales this quarter” represents. Keep quantities and values distinct, and separate new customers from additional purchases by existing customers.

A forecast built from sales opportunities also needs different inputs from a usage-based revenue forecast. Salesforce's training guidance distinguishes these approaches. Salesforce Trailhead.

Record the movements behind the number

For each material component, ask:

• Which customer or clearly defined customer group does it concern?
• Which product, quantity, unit and period are involved?
• Where is the current status recorded?
• Which purchasing, approval or delivery condition remains unresolved?
• When was the last meaningful update?
• Who included the component in the forecast, and on what basis?

A CRM stage label is not sufficient by itself. A proposal stage should be supported by a record that the proposal was shared and that the counterparty's next step is understood. Moving an old opportunity's closing date forward does not create new evidence.

Separate duplication from changes in timing

Suppose an expected additional purchase appears in both an account manager's forecast and a general market-growth allowance. The total may appear to have two sources of support even though the same demand has been counted twice.

Define the boundary between named customer or product items and aggregate growth assumptions. When an order moves into a later period, review both periods together. Do not count the rescheduled item as a new win.

Examine the basis of probability estimates

Assigning a percentage to an opportunity does not establish that the percentage is calibrated. If historical conversion rates are used, check whether they concern comparable products, customers, stage definitions and sales cycles. Make uncertainty explicit when observations are scarce or the process has changed.

Where the data does not support precise percentages, conditions can be more informative: customer approval pending, technical acceptance unresolved or delivery timing unconfirmed. If you prepare scenarios, identify which condition changes which component of the forecast.

Compare results with the forecast version used for the decision

Preserve the forecast from before the period begins, or from the agreed decision date. Comparing actual results only with the latest revised forecast can hide earlier uncertainty. Examine differences in volume, timing, cancellation and scope separately.

A good result in one period does not establish that the method is reliable in every setting. Look for recurring differences concentrated in particular customers or stages. The purpose is to find which assumption needs attention before the next decision, as well as to measure forecast error.

Before a material commitment based on expected sales, Countersign provides a written second opinion on the core assumptions of an existing plan. Briefly describe the plan and your decision date.